The authors compared the survival probabilities of atmospheric particles using theory, simulations, and observation data in Beijing. I find the the comparison with observation data especially valuable, and this work can be nice contribution to NPF studies. However, questions below need to be addressed before the manuscript can be published.

1. Line 183: ‘Evaporation in the cluster formation from sulfuricacid and dimethylamine has been observed to be negligible’. This sentence is somewhat misleading and ‘has been’ is very vague. The evaporation rate is only close to negligible when the base concentration is very high. Jen et al. (2014) listed the nucleation rates in their Table 1. I would rather claim the simulation is a reference in the kinetic limit for comparison with the analytical formulae or observation.
2. Line 270: ‘the loss rates of the clusters can also be considerably affected by the cluster-cluster collisions inaddition to the GRs if the background CS is small and the cluster concentrations are high’. This sentence seems to suggest growth will cause the cluster to be lost, which is incorrect.
3. Figure 2. I would suggest adding simulated rates with cluster-cluster coagulation to the right y axis of figure 2. Since simulation with cluster-cluster coagulation is extensively shown in subsequent figures, this is important information.
4. I find Figure 5 challenging to read as there are too many cases presented in a single figure. I suggest reducing the number of cases presented in this figure. For instance, one fitting method can be presented while the other may go into supplemental.
5. The simulation is extensively compared to the observations. However, I wonder if the simulation is a good representation of atmospheric particle growth: in the simulation there is only one condensing vapor, while in the atmosphere the condensable vapor concentration increases with particle size. This is especially the case for growth between 1.5 nm to 3 nm when the Kelvin effect is very strong. For larger particles it might be alright because the ‘condensable’ concentration is no longer a strong function of particle size; using a single vapor to model the growth rate may coincide with the real atmosphere.  Therefore, how is it justifiable that the simulation is directly comparable to atmospheric observations, especially for Figure 7? (The authors sort of discussed this matter in section 3.4.3 when the uncertainty of GR is discussed, but I wonder why the participation of other species on growth is not explicitly discussed) Overall, the role of the simulation needs to be better defined/clarified.
6. Line 361: what is the meaning of ‘median NPF event’? Median of what? The word ‘median’ is extensively used in section 3.4.1. I found many of them really confusing.
7. Section 3.4.2: The sequence of presentation is non-linear: the authors first briefly discuss Figures 7-9 briefly and then discuss Figures 7 and 8 again with more depth. I wonder if it’s better to just sequentially discuss Figures 7-9.
8. Line 455: I have doubts regarding uncertainty of J₅. The measurement uncertainty decreases with increasing particle size, hence I assume the uncertainly of J_1.5 is larger than that of J₃. Is it possible that J_1.5 is actually systematically underestimated, as a result of which J₃/J_1.5 is larger than reality? From what I know sub 3 nm particles often seem under detected in the observed PSD, does this have an effect?

Technical corrections:

1. The rate constant is missing in the third term on the R.H.S. of Eqn. (8).
2. In the top panel of Figure 6, tick labels are missing on the left y-axis. Also, can the color scale be changed for this plot to make NPF event more visible?

The manuscript deals with survival properties of freshly nucleated particles while they grow and scavenge. Simple parameterizations and an aerosol dynamic model are compared against observations in Beijing, where the high survival rates have puzzled researchers because background aerosol concentrations are so high that the sink they cause should (theoretically) result in lower survival rates than what is observed. The usefulness of the manuscript, from my point of view, lies in the uncertainty analyses presented and the discussions related to potential causes of the above-mentioned discrepancy. The manuscript is definitely suitable for publication in ACP, however, some issues have to be clarified prior to this.

Major comments:

1. Several of the authors here are authors in two other papers/manuscripts, also dealing with the topic of survival rates in Beijing: Cai et al. (ACP 22, p. 11529, 2022) and Cai et al. (ACPD, acp-2022-476, 2022). In the first one of these, the authors state in the conclusions: “The low theoretical survival probabilities of new particles contrasting to the new particle formation events observed at high coagulation sinks should be caused by underestimated growth rates”. In the latter one, the authors claim that experimental survival probability for ‘Beijing-type’ events should be determined by comparing pointwise dN/d(logdp) values and with this approach “the measured and theoretical survival probabilities are on average consistent with each other”. In this manuscript, under review, the analysis is based on studying what kind of uncertainty is needed in the estimated values for GR and CS for the theory/model to match experimental findings.

The three papers/manuscripts thus approach the concept of survival from slightly different angles, and a short discussion (with references) should be written about the differences/similarities in these. And, if possible, some speculative conclusions would be nice about the combined results of these studies, e.g. have we so far analyzed experimental survival incorrectly (as implied in acp-2022-476) or have we analyzed GR somehow incorrectly or are the measurements so uncertain that observational survival results with some reasonable uncertainty range actually agree with theory/modeling.

2. As also detailed aerosol dynamics modeling has been utilized, it would be nice to see some more details about the results. With this I mean, for example, contour (or banana) plots of the growing nucleation mode for an average NPF day.

Minor comments:

3. Page 2, line 46: Here the survival probability is defined in terms of fluxes (“ratio of the formation rates of particles of diameters d1 and d2”) while in the other paper by the authors (acp-2022-476) it is defined in terms of concentrations. Are these the same? If not, which one is correct?
4. Page 2, line 48: Doesn’t GR also depend on particle phase chemistry?
5. Page 5, equation 7: What are the assumptions inherent in equation 7 if there are any? Is CoagS some kind of average in the interval? And, shouldn’t the last term, if written exactly, be GR x n at the upper edge of the interval?
6. Page 6, lines 147-151. When estimating GR using the appearance time method, is some kind of fitting or smoothing applied to the concentration vs. time data of each channel before the peak value is stored?
7. Pages 6-7, description of the model: The description is a little bit confusing. You mention that ACDC is used, but then state that source, loss and evaporation terms are set to zero. Aren’t you then just solving the discrete coagulation equation? Wouldn’t it be clearer to also write it that way? Also, how realistically do you think this describes events in Beijing, when Kelvin- and Raoult-type effects are neglected?
8. Page 8, line 203: how do you calculate the cluster flux past a certain size?
9. Page 10, line 250 and Fig. 1: For a single component system, If C_mon is constant, and Kevin-effect is neglected, condensation-theory tells that GR should be constant in the free-molecular regime, apart from a region at the smallest particle sizes where molecular dimensions also play a role. This means that the result in Case 1 in no surprise. In the other paper by the authors (acp-2022-476), however, the authors apply a size dependent growth rate (increases with size) in their simulations. Why these different approaches when studying essentially the same thing? One important question is also, if the current simulations with constant GR result in a narrowing size distribution when plotted in dN/d(logdp) form in log-scale.
10. Page 10, lines 260-263: if in the high-CS cases the clusters disappear very rapidly, is I then reasonable to assume a constant monomer concentration in the simulations?
11. Page 10, lines 281-282: J1.5 is almost always smaller if determined from eq. 7 than based on fluxes. Do you know why this is so?
12. Page 11, line 288 and fig. 3: Please add base-case values to figure caption.
13. Page 11, line 290 (and many other places): CS/GR is given as unitless? The (SI) unit of CS is 1/s and of GR m/s so how can their ratio be unitless?
14. Page 13, line 361: What is a median event day? How do you determine it?
15. Page 13, line 363: Should it be “…based on Eq. 13 and Eq. 14.” ?
16. Page 14, lines 403-405. Do van der Waals forces affect CoagS also?
17. Page 15, lines 435-438: One thing important to mention about self-coagulation is also that as a process it also effectively decreases the number concentration. If particles of diameter 3 nm grow to 6 nm by self-coagulation alone, their volume increases by a factor of 8 so that their number concentration also would decrease by the same factor.